j FSrSSKrSSKJr SSKJr SSKrSSKJr SSKJ r SSK J r SS K JrJr SS KJr SS KJr SS KJr SS KJrJr SSKJrJr SSKJr SSKJrJ r J!r!J"r" SSK#J$r$ SSK%J&r& SSK'J(r( \"RR"\*5r+\!\"SS\555r,"SS5r-\!\"SS\555r.\!\"SS\555r/"SS\R`5r1"SS \R`5r2"S!S"\R`5r3"S#S$\R`5r4"S%S&\R`5r5"S'S(\R`5r6S)r7SPS*jr8"S+S,\R`5r9S-\RtS.\;S/\Rt4S0jr<SQS1\R`S2\RtS3\RtS4\RtS5\RtS-S6\=S7\=S8\\ 4S9jjr>"S:S;\R`5r?"S<S=\5r@"S>S?\R`5rA"S@SA\R`5rB"SBSC\R`5rC"SDSE\R`5rD"SFSG\R`5rE"SHSI\R`5rF\!"SJSK\55rG\!"SLSM9"SNSO\G55rHSOSK/rIg)RzPyTorch Mimi model.N)Callable) dataclass)nn)initialization)ACT2FN)Cache DynamicCache)!create_sliding_window_causal_mask)GradientCheckpointingLayer)BaseModelOutputWithPast)ROPE_INIT_FUNCTIONSdynamic_rope_update)ALL_ATTENTION_FUNCTIONSPreTrainedModel)Unpack) ModelOutputTransformersKwargsauto_docstringlogging)deprecate_kwarg)maybe_autocast) MimiConfigc\rSrSr%SrSr\RS-\S'Sr \RS-\S'Sr \ S-\S'Sr \ S-\S'Srg) MimiOutput)a audio_codes (`torch.LongTensor` of shape `(batch_size, num_quantizers, codes_length)`, *optional*): Discrete code embeddings computed using `model.encode`. audio_values (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*): Decoded audio values, obtained using the decoder part of Mimi. encoder_past_key_values (`Cache`, *optional*): Pre-computed hidden-states (key and values in the self-attention blocks) that can be used to speed up sequential decoding of the encoder transformer. This typically consists in the `past_key_values` returned by the model at a previous stage of decoding, when `use_cache=True` or `config.use_cache=True`. The model will output the same cache format that is fed as input. If `past_key_values` are used, the user can optionally input only the last `audio_values` or `audio_codes (those that don't have their past key value states given to this model). decoder_past_key_values (`Cache`, *optional*): Pre-computed hidden-states (key and values in the self-attention blocks) that can be used to speed up sequential decoding of the decoder transformer. This typically consists in the `past_key_values` returned by the model at a previous stage of decoding, when `use_cache=True` or `config.use_cache=True`. The model will output the same cache format that is fed as input. If `past_key_values` are used, the user can optionally input only the last `audio_values` or `audio_codes (those that don't have their past key value states given to this model). N audio_codes audio_valuesencoder_past_key_valuesdecoder_past_key_values)__name__ __module__ __qualname____firstlineno____doc__rtorch LongTensor__annotations__r FloatTensorr r r!__static_attributes__r"ڴC:\Users\johnn\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.13_qbz5n2kfra8p0\LocalCache\local-packages\Python313\site-packages\transformers/models/mimi/modeling_mimi.pyrr)sQ.,0K!!D(/-1L%##d*1,0UT\0,0UT\0r-rc \rSrSrSrS\S\\S\\S\\4SjrS\ RS \4S jr S\ RS \4S jr S r g )MimiConv1dPaddingCacheIa{ Padding cache for MimiConv1d causal convolutions in order to support streaming via cache padding. See: https://huggingface.co/papers/2005.06720 & https://huggingface.co/papers/2204.07064 A padding cache is a list of cached partial hidden states for each convolution layer. Hidden states are cached from the previous call to the MimiConv1d forward pass, given the padding size. num_layersper_layer_paddingper_layer_padding_modeper_layer_in_channelsc[U5[U5[U51n[U5S:wdUR5U:wa[SUS35eX lX0lX@lS/U-Ulg)NrzExpected `num_layers` (zU) values in `per_layer_padding`, `per_layer_padding_mode` and `per_layer_in_channels`)lenpop ValueErrorr3r4r5 padding_cache)selfr2r3r4r5from_args_num_layerss r.__init__MimiConv1dPaddingCache.__init__Rs!$$5 6